基于误分类代价的粗糙模糊集近似集

粗糙模糊集的近似集提供了如何利用已知的信息粒来近似描述模糊知识的方法,但在构建近似集时并没有考虑误分类代价这一实际因素.针对此问题,从误分类代价的角度,提出了粗糙模糊集的近似表示R(X),并揭示了多粒度知识空间中不确定性域对应的误分类代价随着粒度细化的变化规律.实验结果表明,R(X),R(X)和R(X)分别作为X的近似集时,R(X)产生的误分类代价最小,在一定程度上反映了R(X)作为近似集时的优势.

Approximation set of rough fuzzy set based on misclassification cost

Although the approximation set of rough fuzzy set provides a method to approximately describe fuzzy knowledge by using existing information granules, the misclassification cost is not considered when the approximation set is constructed. To solve this problem, this paper proposes the approximation set of rough fuzzy set R(X) based on misclassification cost. The change rules of the misclassification cost of uncertain region in the multi-granularity knowledge spaces are revealed. The experimental results demonstrate that when R(X), R (X) and R (X) are the approximation set of X, the misclassification cost of R(X) is minimum, which reflects the advantages of R(X).